Data‐driven prediction and multi‐objective emission control in co‐firing systems using support vector regression and genetic algorithm

Abstract Current study investigates the emission and combustion performance of a tangentially fired boiler co‐firing coal with municipal solid waste (MSW) and biomass using support vector regression (SVR) modeling integrated with a genetic algorithm (GA) Pareto optimization and proposes a novel hybrid modeling framework. Experimental data were collected across a 60–110 MW load range, and SVR models were developed to capture nonlinear behavior in SO x , NO x , suspended particulate matter, flue gas temperature, and flame intensity. The GA Pareto method was then applied to identify optimal non‐dominated operating points for emission minimization and combustion enhancement. The results show that co‐firing 10% MSW reduces SO x emissions by 29–32%, while 10% biomass reduces SO x by 10–13% relative to optimized coal firing in the stable range. For NO x , biomass blending shows a 2–4% reduction, whereas MSW shows a slight increase due to higher fuel‐bound nitrogen. Combustion performance improved significantly under MSW and biomass fuels and increased flue gas temperatures to 820°C and 830°C, compared to 700°C for coal only, while flame intensity peaked at 45.5 for MSW, 6% above coal. For all responses, the GA Pareto optima consistently clustered in the 90–105 MW range, confirming this as the most favorable operational band for balancing emission control and combustion stability. Overall, the hybrid SVR–GA methodology proved highly effective in identifying realistic, multi‐objective operating strategies for waste and biomass co‐firing in boiler. The combined SVR–GA approach thus provides a validated and intelligent model for co‐firing decision‐making.